The landcover of the northern floodplain around the Tonle Sap Lake involves the various vegetations, lacustrine lands, as well as settlements. In order to understand the contribution of landcover in this area for agricultural, piscicultural activity, and environmental protection, landcover classes should be classified by using remote sensing data. The aim of this study is to increase distinction between landcover classes for classification purpose. To improve the feature texture for pre-classification data, the ALOS PALSAR is fused with ASTER data. Both data are acquired in dry season in which the vegetation is little influenced by flooding. The fused data is created by injecting the feature texture of ALOS PALSAR into ASTER data. However, spectral character is distorted due to mixed spectrum. This is reduced by choosing optimal fused algorithm. The ten landcover classes are selected as signatures to classify and calculate confusion matrixes. Those confusion matrixes reveal that the distinction between the landcover classes in fused data is better than that in ASTER data.